{
  "aggregates": [
    {
      "all_in_output_estimate_total_usd": null,
      "all_in_output_estimate_upper_bound_total_usd": null,
      "cell": "glm-openrouter",
      "cost_per_success_usd": 0.146551,
      "costed_runs": 3,
      "image_output_estimate_total_usd": null,
      "image_output_estimate_upper_bound_total_usd": null,
      "lane": "hermes",
      "median_cost_usd": 0.150947,
      "median_wall_seconds": 516.232,
      "runs": 3,
      "success_rate": 1.0,
      "successes": 3,
      "total_cost_usd": 0.439653,
      "total_wall_seconds": 1730.516,
      "workload": "eventforge"
    },
    {
      "all_in_output_estimate_total_usd": null,
      "all_in_output_estimate_upper_bound_total_usd": null,
      "cell": "glm-openrouter",
      "cost_per_success_usd": 0.05023,
      "costed_runs": 3,
      "image_output_estimate_total_usd": null,
      "image_output_estimate_upper_bound_total_usd": null,
      "lane": "pft",
      "median_cost_usd": 0.050411,
      "median_wall_seconds": 265.688,
      "runs": 3,
      "success_rate": 1.0,
      "successes": 3,
      "total_cost_usd": 0.150689,
      "total_wall_seconds": 1180.345,
      "workload": "eventforge"
    },
    {
      "all_in_output_estimate_total_usd": null,
      "all_in_output_estimate_upper_bound_total_usd": null,
      "cell": "kimi-openrouter",
      "cost_per_success_usd": 0.802998,
      "costed_runs": 3,
      "image_output_estimate_total_usd": null,
      "image_output_estimate_upper_bound_total_usd": null,
      "lane": "hermes",
      "median_cost_usd": 0.356091,
      "median_wall_seconds": 314.765,
      "runs": 3,
      "success_rate": 0.6667,
      "successes": 2,
      "total_cost_usd": 1.605997,
      "total_wall_seconds": 1792.318,
      "workload": "eventforge"
    },
    {
      "all_in_output_estimate_total_usd": null,
      "all_in_output_estimate_upper_bound_total_usd": null,
      "cell": "kimi-openrouter",
      "cost_per_success_usd": 0.380092,
      "costed_runs": 3,
      "image_output_estimate_total_usd": null,
      "image_output_estimate_upper_bound_total_usd": null,
      "lane": "pft",
      "median_cost_usd": 0.348651,
      "median_wall_seconds": 582.386,
      "runs": 3,
      "success_rate": 1.0,
      "successes": 3,
      "total_cost_usd": 1.140275,
      "total_wall_seconds": 1571.644,
      "workload": "eventforge"
    },
    {
      "all_in_output_estimate_total_usd": null,
      "all_in_output_estimate_upper_bound_total_usd": null,
      "cell": "glm-openrouter",
      "cost_per_success_usd": 0.341135,
      "costed_runs": 3,
      "image_output_estimate_total_usd": null,
      "image_output_estimate_upper_bound_total_usd": null,
      "lane": "hermes",
      "median_cost_usd": 0.323866,
      "median_wall_seconds": 288.692,
      "runs": 3,
      "success_rate": 1.0,
      "successes": 3,
      "total_cost_usd": 1.023405,
      "total_wall_seconds": 1063.555,
      "workload": "queuecraft"
    },
    {
      "all_in_output_estimate_total_usd": null,
      "all_in_output_estimate_upper_bound_total_usd": null,
      "cell": "glm-openrouter",
      "cost_per_success_usd": 0.321662,
      "costed_runs": 3,
      "image_output_estimate_total_usd": null,
      "image_output_estimate_upper_bound_total_usd": null,
      "lane": "pft",
      "median_cost_usd": 0.272193,
      "median_wall_seconds": 377.436,
      "runs": 3,
      "success_rate": 1.0,
      "successes": 3,
      "total_cost_usd": 0.964986,
      "total_wall_seconds": 957.643,
      "workload": "queuecraft"
    },
    {
      "all_in_output_estimate_total_usd": null,
      "all_in_output_estimate_upper_bound_total_usd": null,
      "cell": "glm-vercel",
      "cost_per_success_usd": 0.579976,
      "costed_runs": 3,
      "image_output_estimate_total_usd": null,
      "image_output_estimate_upper_bound_total_usd": null,
      "lane": "hermes",
      "median_cost_usd": 0.572873,
      "median_wall_seconds": 304.641,
      "runs": 3,
      "success_rate": 1.0,
      "successes": 3,
      "total_cost_usd": 1.739928,
      "total_wall_seconds": 790.371,
      "workload": "queuecraft"
    },
    {
      "all_in_output_estimate_total_usd": null,
      "all_in_output_estimate_upper_bound_total_usd": null,
      "cell": "glm-vercel",
      "cost_per_success_usd": 0.349323,
      "costed_runs": 3,
      "image_output_estimate_total_usd": null,
      "image_output_estimate_upper_bound_total_usd": null,
      "lane": "pft",
      "median_cost_usd": 0.359863,
      "median_wall_seconds": 158.72,
      "runs": 3,
      "success_rate": 1.0,
      "successes": 3,
      "total_cost_usd": 1.04797,
      "total_wall_seconds": 476.914,
      "workload": "queuecraft"
    },
    {
      "all_in_output_estimate_total_usd": null,
      "all_in_output_estimate_upper_bound_total_usd": null,
      "cell": "kimi-openrouter",
      "cost_per_success_usd": 1.601043,
      "costed_runs": 3,
      "image_output_estimate_total_usd": null,
      "image_output_estimate_upper_bound_total_usd": null,
      "lane": "hermes",
      "median_cost_usd": 1.549447,
      "median_wall_seconds": 705.385,
      "runs": 3,
      "success_rate": 1.0,
      "successes": 3,
      "total_cost_usd": 4.803129,
      "total_wall_seconds": 1844.472,
      "workload": "queuecraft"
    },
    {
      "all_in_output_estimate_total_usd": null,
      "all_in_output_estimate_upper_bound_total_usd": null,
      "cell": "kimi-openrouter",
      "cost_per_success_usd": 0.305103,
      "costed_runs": 3,
      "image_output_estimate_total_usd": null,
      "image_output_estimate_upper_bound_total_usd": null,
      "lane": "pft",
      "median_cost_usd": 0.34042,
      "median_wall_seconds": 143.329,
      "runs": 3,
      "success_rate": 1.0,
      "successes": 3,
      "total_cost_usd": 0.91531,
      "total_wall_seconds": 435.786,
      "workload": "queuecraft"
    },
    {
      "all_in_output_estimate_total_usd": null,
      "all_in_output_estimate_upper_bound_total_usd": null,
      "cell": "glm-openrouter",
      "cost_per_success_usd": 3.741205,
      "costed_runs": 3,
      "image_output_estimate_total_usd": null,
      "image_output_estimate_upper_bound_total_usd": null,
      "lane": "hermes",
      "median_cost_usd": 1.179643,
      "median_wall_seconds": 1456.798,
      "runs": 3,
      "success_rate": 0.3333,
      "successes": 1,
      "total_cost_usd": 3.741205,
      "total_wall_seconds": 5366.347,
      "workload": "visual_site"
    },
    {
      "all_in_output_estimate_total_usd": 4.230076,
      "all_in_output_estimate_upper_bound_total_usd": 4.230076,
      "cell": "glm-openrouter",
      "cost_per_success_usd": 0.519692,
      "costed_runs": 3,
      "image_output_estimate_total_usd": 2.671,
      "image_output_estimate_upper_bound_total_usd": 2.671,
      "lane": "pft",
      "median_cost_usd": 0.374547,
      "median_wall_seconds": 1497.89,
      "runs": 3,
      "success_rate": 1.0,
      "successes": 3,
      "total_cost_usd": 1.559076,
      "total_wall_seconds": 4554.221,
      "workload": "visual_site"
    },
    {
      "all_in_output_estimate_total_usd": 9.549964,
      "all_in_output_estimate_upper_bound_total_usd": 9.549964,
      "cell": "kimi-openrouter",
      "cost_per_success_usd": 2.523321,
      "costed_runs": 3,
      "image_output_estimate_total_usd": 1.98,
      "image_output_estimate_upper_bound_total_usd": 1.98,
      "lane": "hermes",
      "median_cost_usd": 2.172302,
      "median_wall_seconds": 1840.704,
      "runs": 3,
      "success_rate": 1.0,
      "successes": 3,
      "total_cost_usd": 7.569964,
      "total_wall_seconds": 5622.937,
      "workload": "visual_site"
    },
    {
      "all_in_output_estimate_total_usd": 6.661211,
      "all_in_output_estimate_upper_bound_total_usd": 6.661211,
      "cell": "kimi-openrouter",
      "cost_per_success_usd": 1.74907,
      "costed_runs": 3,
      "image_output_estimate_total_usd": 1.414,
      "image_output_estimate_upper_bound_total_usd": 1.414,
      "lane": "pft",
      "median_cost_usd": 1.686337,
      "median_wall_seconds": 1546.911,
      "runs": 3,
      "success_rate": 1.0,
      "successes": 3,
      "total_cost_usd": 5.247211,
      "total_wall_seconds": 4439.645,
      "workload": "visual_site"
    },
    {
      "all_in_output_estimate_total_usd": 21.733316,
      "all_in_output_estimate_upper_bound_total_usd": 21.733316,
      "cell": "opus",
      "cost_per_success_usd": 10.055158,
      "costed_runs": 3,
      "image_output_estimate_total_usd": 1.623,
      "image_output_estimate_upper_bound_total_usd": 1.623,
      "lane": "cc",
      "median_cost_usd": 6.642805,
      "median_wall_seconds": 1689.07,
      "runs": 3,
      "success_rate": 0.6667,
      "successes": 2,
      "total_cost_usd": 20.110315,
      "total_wall_seconds": 5337.666,
      "workload": "visual_site"
    },
    {
      "all_in_output_estimate_total_usd": 10.344594,
      "all_in_output_estimate_upper_bound_total_usd": 10.344594,
      "cell": "opus",
      "cost_per_success_usd": 2.891865,
      "costed_runs": 3,
      "image_output_estimate_total_usd": 1.669,
      "image_output_estimate_upper_bound_total_usd": 1.669,
      "lane": "pft",
      "median_cost_usd": 2.85552,
      "median_wall_seconds": 988.026,
      "runs": 3,
      "success_rate": 1.0,
      "successes": 3,
      "total_cost_usd": 8.675595,
      "total_wall_seconds": 3061.889,
      "workload": "visual_site"
    }
  ],
  "comparisons": [
    {
      "cell": "glm-openrouter",
      "directional_speed_waves": [
        true,
        true,
        true
      ],
      "opponent": "hermes",
      "opponent_successes": 3,
      "pft_all_in_output_estimate_advantage_x_total": null,
      "pft_all_in_output_estimate_savings_percent": null,
      "pft_cost_advantage_x_total": 2.9176,
      "pft_cost_savings_percent": 65.73,
      "pft_speedup_x_total": 1.4661,
      "pft_successes": 3,
      "speed_result": "consistent_pft",
      "workload": "eventforge"
    },
    {
      "cell": "kimi-openrouter",
      "directional_speed_waves": [
        false,
        true
      ],
      "opponent": "hermes",
      "opponent_successes": 2,
      "pft_all_in_output_estimate_advantage_x_total": null,
      "pft_all_in_output_estimate_savings_percent": null,
      "pft_cost_advantage_x_total": 1.4084,
      "pft_cost_savings_percent": 29.0,
      "pft_speedup_x_total": 1.1404,
      "pft_successes": 3,
      "speed_result": "mixed_or_incomplete",
      "workload": "eventforge"
    },
    {
      "cell": "glm-openrouter",
      "directional_speed_waves": [
        true,
        false,
        false
      ],
      "opponent": "hermes",
      "opponent_successes": 3,
      "pft_all_in_output_estimate_advantage_x_total": null,
      "pft_all_in_output_estimate_savings_percent": null,
      "pft_cost_advantage_x_total": 1.0605,
      "pft_cost_savings_percent": 5.71,
      "pft_speedup_x_total": 1.1106,
      "pft_successes": 3,
      "speed_result": "mixed_or_incomplete",
      "workload": "queuecraft"
    },
    {
      "cell": "glm-vercel",
      "directional_speed_waves": [
        true,
        true,
        true
      ],
      "opponent": "hermes",
      "opponent_successes": 3,
      "pft_all_in_output_estimate_advantage_x_total": null,
      "pft_all_in_output_estimate_savings_percent": null,
      "pft_cost_advantage_x_total": 1.6603,
      "pft_cost_savings_percent": 39.77,
      "pft_speedup_x_total": 1.6573,
      "pft_successes": 3,
      "speed_result": "consistent_pft",
      "workload": "queuecraft"
    },
    {
      "cell": "kimi-openrouter",
      "directional_speed_waves": [
        true,
        true,
        true
      ],
      "opponent": "hermes",
      "opponent_successes": 3,
      "pft_all_in_output_estimate_advantage_x_total": null,
      "pft_all_in_output_estimate_savings_percent": null,
      "pft_cost_advantage_x_total": 5.2475,
      "pft_cost_savings_percent": 80.94,
      "pft_speedup_x_total": 4.2325,
      "pft_successes": 3,
      "speed_result": "consistent_pft",
      "workload": "queuecraft"
    },
    {
      "cell": "glm-openrouter",
      "directional_speed_waves": [
        false
      ],
      "opponent": "hermes",
      "opponent_successes": 1,
      "pft_all_in_output_estimate_advantage_x_total": null,
      "pft_all_in_output_estimate_savings_percent": null,
      "pft_cost_advantage_x_total": 2.3996,
      "pft_cost_savings_percent": 58.33,
      "pft_speedup_x_total": 1.1783,
      "pft_successes": 3,
      "speed_result": "mixed_or_incomplete",
      "workload": "visual_site"
    },
    {
      "cell": "kimi-openrouter",
      "directional_speed_waves": [
        true,
        false,
        true
      ],
      "opponent": "hermes",
      "opponent_successes": 3,
      "pft_all_in_output_estimate_advantage_x_total": 1.4337,
      "pft_all_in_output_estimate_savings_percent": 30.25,
      "pft_cost_advantage_x_total": 1.4427,
      "pft_cost_savings_percent": 30.68,
      "pft_speedup_x_total": 1.2665,
      "pft_successes": 3,
      "speed_result": "mixed_or_incomplete",
      "workload": "visual_site"
    },
    {
      "cell": "opus",
      "directional_speed_waves": [
        true,
        true
      ],
      "opponent": "cc",
      "opponent_successes": 2,
      "pft_all_in_output_estimate_advantage_x_total": 2.1009,
      "pft_all_in_output_estimate_savings_percent": 52.4,
      "pft_cost_advantage_x_total": 2.318,
      "pft_cost_savings_percent": 56.86,
      "pft_speedup_x_total": 1.7433,
      "pft_successes": 3,
      "speed_result": "mixed_or_incomplete",
      "workload": "visual_site"
    }
  ],
  "conformance_exclusions": [
    {
      "artifact": "visual/results/glm-openrouter/hermes/wave2/agent_run.json",
      "cell": "glm-openrouter",
      "lane": "hermes",
      "reasons": [
        "Hermes terminal isolation removed the supplied OPENAI_API_KEY.",
        "The agent generated final assets through the ChatGPT Codex OAuth backend using gpt-image-2-codex rather than the required OpenAI Image API gpt-image-2 route."
      ],
      "wave": 2,
      "workload": "visual_site"
    },
    {
      "artifact": "visual/results/glm-openrouter/hermes/wave3/agent_run.json",
      "cell": "glm-openrouter",
      "lane": "hermes",
      "reasons": [
        "Hermes terminal isolation removed the supplied OPENAI_API_KEY.",
        "The agent generated final assets through a discovered Vercel AI Gateway credential rather than the matched direct OpenAI Image API route."
      ],
      "wave": 3,
      "workload": "visual_site"
    }
  ],
  "judge_overhead": {
    "cache_write_tokens": 618624,
    "cached_input_tokens": 0,
    "estimated_cost_usd": 6.11742,
    "input_tokens": 618678,
    "output_tokens": 75025,
    "passes": 18,
    "source": "GPT-5.6-Sol response usage reconstructed with official standard short-context prices"
  },
  "raw_runs": [
    {
      "artifact": "visual/results/glm-openrouter/hermes/wave1/agent_run.json",
      "cell": "glm-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": 1.931643,
      "contestant_plus_image_output_estimate_usd": 1.720643,
      "cost_usd": 1.179643,
      "final_image_count": 3,
      "image_cost_evidence": "attempt_audit",
      "image_count": 3,
      "image_output_estimate_upper_bound_usd": 0.752,
      "image_output_estimate_usd": 0.541,
      "lane": "hermes",
      "model": "z-ai/glm-5.2",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 1,
      "wall_seconds": 1288.428,
      "wave": 1,
      "workload": "visual_site"
    },
    {
      "artifact": "visual/results/glm-openrouter/hermes/wave2/agent_run.json",
      "cell": "glm-openrouter",
      "conformance_reasons": [
        "Hermes terminal isolation removed the supplied OPENAI_API_KEY.",
        "The agent generated final assets through the ChatGPT Codex OAuth backend using gpt-image-2-codex rather than the required OpenAI Image API gpt-image-2 route."
      ],
      "conformance_valid": false,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 1.003355,
      "final_image_count": 3,
      "image_cost_evidence": "attempt_audit_unsupported_route_or_price",
      "image_count": 5,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "hermes",
      "model": "z-ai/glm-5.2",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": false,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 1456.798,
      "wave": 2,
      "workload": "visual_site"
    },
    {
      "artifact": "visual/results/glm-openrouter/hermes/wave3/agent_run.json",
      "cell": "glm-openrouter",
      "conformance_reasons": [
        "Hermes terminal isolation removed the supplied OPENAI_API_KEY.",
        "The agent generated final assets through a discovered Vercel AI Gateway credential rather than the matched direct OpenAI Image API route."
      ],
      "conformance_valid": false,
      "contestant_plus_image_output_estimate_upper_bound_usd": 3.072207,
      "contestant_plus_image_output_estimate_usd": 2.990207,
      "cost_usd": 1.558207,
      "final_image_count": 3,
      "image_cost_evidence": "attempt_audit",
      "image_count": 15,
      "image_output_estimate_upper_bound_usd": 1.514,
      "image_output_estimate_usd": 1.432,
      "lane": "hermes",
      "model": "z-ai/glm-5.2",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": false,
      "timed_out": false,
      "uncertain_image_count": 2,
      "wall_seconds": 2621.121,
      "wave": 3,
      "workload": "visual_site"
    },
    {
      "artifact": "visual/results/glm-openrouter/pft/wave1/agent_run.json",
      "cell": "glm-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": 1.706904,
      "contestant_plus_image_output_estimate_usd": 1.706904,
      "cost_usd": 0.328904,
      "final_image_count": 3,
      "image_cost_evidence": "attempt_audit",
      "image_count": 10,
      "image_output_estimate_upper_bound_usd": 1.378,
      "image_output_estimate_usd": 1.378,
      "lane": "pft",
      "model": "z-ai/glm-5.2",
      "pft_call_diagnostics": {
        "calls": 36,
        "finish_reasons": {
          "ok": 36
        },
        "providers": {
          "StreamLake": 36
        },
        "retry_attempts": 0
      },
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 1836.331,
      "wave": 1,
      "workload": "visual_site"
    },
    {
      "artifact": "visual/results/glm-openrouter/pft/wave2/agent_run.json",
      "cell": "glm-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": 1.561625,
      "contestant_plus_image_output_estimate_usd": 1.561625,
      "cost_usd": 0.855625,
      "final_image_count": 3,
      "image_cost_evidence": "attempt_audit",
      "image_count": 4,
      "image_output_estimate_upper_bound_usd": 0.706,
      "image_output_estimate_usd": 0.706,
      "lane": "pft",
      "model": "z-ai/glm-5.2",
      "pft_call_diagnostics": {
        "calls": 67,
        "finish_reasons": {
          "ok": 67
        },
        "providers": {
          "Alibaba": 37,
          "Novita": 8,
          "StreamLake": 22
        },
        "retry_attempts": 0
      },
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 1497.89,
      "wave": 2,
      "workload": "visual_site"
    },
    {
      "artifact": "visual/results/glm-openrouter/pft/wave3/agent_run.json",
      "cell": "glm-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": 0.961547,
      "contestant_plus_image_output_estimate_usd": 0.961547,
      "cost_usd": 0.374547,
      "final_image_count": 3,
      "image_cost_evidence": "attempt_audit",
      "image_count": 3,
      "image_output_estimate_upper_bound_usd": 0.587,
      "image_output_estimate_usd": 0.587,
      "lane": "pft",
      "model": "z-ai/glm-5.2",
      "pft_call_diagnostics": {
        "calls": 40,
        "finish_reasons": {
          "ok": 40
        },
        "providers": {
          "StreamLake": 40
        },
        "retry_attempts": 0
      },
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 1220.0,
      "wave": 3,
      "workload": "visual_site"
    },
    {
      "artifact": "visual/results/kimi-openrouter/hermes/wave1/agent_run.json",
      "cell": "kimi-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": 4.455852,
      "contestant_plus_image_output_estimate_usd": 4.455852,
      "cost_usd": 3.465852,
      "final_image_count": 3,
      "image_cost_evidence": "attempt_audit",
      "image_count": 6,
      "image_output_estimate_upper_bound_usd": 0.99,
      "image_output_estimate_usd": 0.99,
      "lane": "hermes",
      "model": "moonshotai/kimi-k3",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 1840.704,
      "wave": 1,
      "workload": "visual_site"
    },
    {
      "artifact": "visual/results/kimi-openrouter/hermes/wave2/agent_run.json",
      "cell": "kimi-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": 2.667302,
      "contestant_plus_image_output_estimate_usd": 2.667302,
      "cost_usd": 2.172302,
      "final_image_count": 3,
      "image_cost_evidence": "attempt_audit",
      "image_count": 3,
      "image_output_estimate_upper_bound_usd": 0.495,
      "image_output_estimate_usd": 0.495,
      "lane": "hermes",
      "model": "moonshotai/kimi-k3",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 1388.9,
      "wave": 2,
      "workload": "visual_site"
    },
    {
      "artifact": "visual/results/kimi-openrouter/hermes/wave3/agent_run.json",
      "cell": "kimi-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": 2.42681,
      "contestant_plus_image_output_estimate_usd": 2.42681,
      "cost_usd": 1.93181,
      "final_image_count": 3,
      "image_cost_evidence": "attempt_audit",
      "image_count": 3,
      "image_output_estimate_upper_bound_usd": 0.495,
      "image_output_estimate_usd": 0.495,
      "lane": "hermes",
      "model": "moonshotai/kimi-k3",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 2393.333,
      "wave": 3,
      "workload": "visual_site"
    },
    {
      "artifact": "visual/results/kimi-openrouter/pft/wave1/agent_run.json",
      "cell": "kimi-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": 2.181337,
      "contestant_plus_image_output_estimate_usd": 2.181337,
      "cost_usd": 1.686337,
      "final_image_count": 3,
      "image_cost_evidence": "attempt_audit",
      "image_count": 3,
      "image_output_estimate_upper_bound_usd": 0.495,
      "image_output_estimate_usd": 0.495,
      "lane": "pft",
      "model": "moonshotai/kimi-k3",
      "pft_call_diagnostics": {
        "calls": 42,
        "finish_reasons": {
          "ok": 42
        },
        "providers": {
          "Fireworks": 16,
          "Moonshot AI": 26
        },
        "retry_attempts": 0
      },
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 1292.03,
      "wave": 1,
      "workload": "visual_site"
    },
    {
      "artifact": "visual/results/kimi-openrouter/pft/wave2/agent_run.json",
      "cell": "kimi-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": 1.88888,
      "contestant_plus_image_output_estimate_usd": 1.88888,
      "cost_usd": 1.46488,
      "final_image_count": 3,
      "image_cost_evidence": "attempt_audit",
      "image_count": 4,
      "image_output_estimate_upper_bound_usd": 0.424,
      "image_output_estimate_usd": 0.424,
      "lane": "pft",
      "model": "moonshotai/kimi-k3",
      "pft_call_diagnostics": {
        "calls": 40,
        "finish_reasons": {
          "ok": 40
        },
        "providers": {
          "Moonshot AI": 40
        },
        "retry_attempts": 0
      },
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 1546.911,
      "wave": 2,
      "workload": "visual_site"
    },
    {
      "artifact": "visual/results/kimi-openrouter/pft/wave3/agent_run.json",
      "cell": "kimi-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": 2.590994,
      "contestant_plus_image_output_estimate_usd": 2.590994,
      "cost_usd": 2.095994,
      "final_image_count": 3,
      "image_cost_evidence": "attempt_audit",
      "image_count": 3,
      "image_output_estimate_upper_bound_usd": 0.495,
      "image_output_estimate_usd": 0.495,
      "lane": "pft",
      "model": "moonshotai/kimi-k3",
      "pft_call_diagnostics": {
        "calls": 66,
        "finish_reasons": {
          "ok": 65,
          "stream closed before chat completions finished": 1
        },
        "providers": {
          "Fireworks": 23,
          "Moonshot AI": 43
        },
        "retry_attempts": 1
      },
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 1600.704,
      "wave": 3,
      "workload": "visual_site"
    },
    {
      "artifact": "visual/results/opus/cc/wave1/agent_run.json",
      "cell": "opus",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": 5.627564,
      "contestant_plus_image_output_estimate_usd": 5.627564,
      "cost_usd": 5.08656375,
      "final_image_count": 3,
      "image_cost_evidence": "final_manifest_only",
      "image_count": 3,
      "image_output_estimate_upper_bound_usd": 0.541,
      "image_output_estimate_usd": 0.541,
      "lane": "cc",
      "model": "claude-opus-5",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 1420.029,
      "wave": 1,
      "workload": "visual_site"
    },
    {
      "artifact": "visual/results/opus/cc/wave2/agent_run.json",
      "cell": "opus",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": 7.183806,
      "contestant_plus_image_output_estimate_usd": 7.183806,
      "cost_usd": 6.6428055,
      "final_image_count": 3,
      "image_cost_evidence": "final_manifest_only",
      "image_count": 3,
      "image_output_estimate_upper_bound_usd": 0.541,
      "image_output_estimate_usd": 0.541,
      "lane": "cc",
      "model": "claude-opus-5",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 1689.07,
      "wave": 2,
      "workload": "visual_site"
    },
    {
      "artifact": "visual/results/opus/cc/wave3/agent_run.json",
      "cell": "opus",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": 8.921946,
      "contestant_plus_image_output_estimate_usd": 8.921946,
      "cost_usd": 8.380946,
      "final_image_count": 3,
      "image_cost_evidence": "final_manifest_only",
      "image_count": 3,
      "image_output_estimate_upper_bound_usd": 0.541,
      "image_output_estimate_usd": 0.541,
      "lane": "cc",
      "model": "claude-opus-5",
      "pft_call_diagnostics": null,
      "returncode": 1,
      "success": false,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 2228.567,
      "wave": 3,
      "workload": "visual_site"
    },
    {
      "artifact": "visual/results/opus/pft/wave1/agent_run.json",
      "cell": "opus",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": 3.39652,
      "contestant_plus_image_output_estimate_usd": 3.39652,
      "cost_usd": 2.8555205,
      "final_image_count": 3,
      "image_cost_evidence": "final_manifest_only",
      "image_count": 3,
      "image_output_estimate_upper_bound_usd": 0.541,
      "image_output_estimate_usd": 0.541,
      "lane": "pft",
      "model": "claude-opus-5",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 954.982,
      "wave": 1,
      "workload": "visual_site"
    },
    {
      "artifact": "visual/results/opus/pft/wave2/agent_run.json",
      "cell": "opus",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": 3.588637,
      "contestant_plus_image_output_estimate_usd": 3.588637,
      "cost_usd": 3.00163725,
      "final_image_count": 3,
      "image_cost_evidence": "final_manifest_only",
      "image_count": 3,
      "image_output_estimate_upper_bound_usd": 0.587,
      "image_output_estimate_usd": 0.587,
      "lane": "pft",
      "model": "claude-opus-5",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 1118.881,
      "wave": 2,
      "workload": "visual_site"
    },
    {
      "artifact": "visual/results/opus/pft/wave3/agent_run.json",
      "cell": "opus",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": 3.359437,
      "contestant_plus_image_output_estimate_usd": 3.359437,
      "cost_usd": 2.81843725,
      "final_image_count": 3,
      "image_cost_evidence": "final_manifest_only",
      "image_count": 3,
      "image_output_estimate_upper_bound_usd": 0.541,
      "image_output_estimate_usd": 0.541,
      "lane": "pft",
      "model": "claude-opus-5",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 988.026,
      "wave": 3,
      "workload": "visual_site"
    },
    {
      "artifact": "deterministic/eventforge/glm-openrouter/wave1/openrouter/results/hermes/wave1/agent_run.json",
      "cell": "glm-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.104977,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "hermes",
      "model": "z-ai/glm-5.2",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 463.925,
      "wave": 1,
      "workload": "eventforge"
    },
    {
      "artifact": "deterministic/eventforge/glm-openrouter/wave1/openrouter/results/pft/wave1/agent_run.json",
      "cell": "glm-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.050411,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "pft",
      "model": "z-ai/glm-5.2",
      "pft_call_diagnostics": {
        "calls": 6,
        "finish_reasons": {
          "ok": 6
        },
        "providers": {
          "Sail Research": 2,
          "Together": 4
        },
        "retry_attempts": 0
      },
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 259.933,
      "wave": 1,
      "workload": "eventforge"
    },
    {
      "artifact": "deterministic/eventforge/glm-openrouter/wave2/openrouter/results/hermes/wave2/agent_run.json",
      "cell": "glm-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.150947,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "hermes",
      "model": "z-ai/glm-5.2",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 750.359,
      "wave": 2,
      "workload": "eventforge"
    },
    {
      "artifact": "deterministic/eventforge/glm-openrouter/wave2/openrouter/results/pft/wave2/agent_run.json",
      "cell": "glm-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.078456,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "pft",
      "model": "z-ai/glm-5.2",
      "pft_call_diagnostics": {
        "calls": 8,
        "finish_reasons": {
          "ok": 8
        },
        "providers": {
          "Friendli": 8
        },
        "retry_attempts": 0
      },
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 654.724,
      "wave": 2,
      "workload": "eventforge"
    },
    {
      "artifact": "deterministic/eventforge/glm-openrouter/wave3/openrouter/results/hermes/wave3/agent_run.json",
      "cell": "glm-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.183729,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "hermes",
      "model": "z-ai/glm-5.2",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 516.232,
      "wave": 3,
      "workload": "eventforge"
    },
    {
      "artifact": "deterministic/eventforge/glm-openrouter/wave3/openrouter/results/pft/wave3/agent_run.json",
      "cell": "glm-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.021822,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "pft",
      "model": "z-ai/glm-5.2",
      "pft_call_diagnostics": {
        "calls": 7,
        "finish_reasons": {
          "ok": 7
        },
        "providers": {
          "AtlasCloud": 3,
          "Decart": 1,
          "Fireworks": 1,
          "Inceptron": 2
        },
        "retry_attempts": 0
      },
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 265.688,
      "wave": 3,
      "workload": "eventforge"
    },
    {
      "artifact": "deterministic/eventforge/kimi-openrouter/wave1/openrouter/results/hermes/wave1/agent_run.json",
      "cell": "kimi-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.356091,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "hermes",
      "model": "moonshotai/kimi-k3",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 277.521,
      "wave": 1,
      "workload": "eventforge"
    },
    {
      "artifact": "deterministic/eventforge/kimi-openrouter/wave1/openrouter/results/pft/wave1/agent_run.json",
      "cell": "kimi-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.731491,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "pft",
      "model": "moonshotai/kimi-k3",
      "pft_call_diagnostics": {
        "calls": 16,
        "finish_reasons": {
          "ok": 16
        },
        "providers": {
          "Fireworks": 10,
          "Moonshot AI": 6
        },
        "retry_attempts": 0
      },
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 710.7,
      "wave": 1,
      "workload": "eventforge"
    },
    {
      "artifact": "deterministic/eventforge/kimi-openrouter/wave2/openrouter/results/hermes/wave2/agent_run.json",
      "cell": "kimi-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.986039,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "hermes",
      "model": "moonshotai/kimi-k3",
      "pft_call_diagnostics": null,
      "returncode": -15,
      "success": false,
      "timed_out": true,
      "uncertain_image_count": 0,
      "wall_seconds": 1200.032,
      "wave": 2,
      "workload": "eventforge"
    },
    {
      "artifact": "deterministic/eventforge/kimi-openrouter/wave2/openrouter/results/pft/wave2/agent_run.json",
      "cell": "kimi-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.348651,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "pft",
      "model": "moonshotai/kimi-k3",
      "pft_call_diagnostics": {
        "calls": 9,
        "finish_reasons": {
          "ok": 9
        },
        "providers": {
          "Fireworks": 1,
          "Moonshot AI": 8
        },
        "retry_attempts": 0
      },
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 582.386,
      "wave": 2,
      "workload": "eventforge"
    },
    {
      "artifact": "deterministic/eventforge/kimi-openrouter/wave3/openrouter/results/hermes/wave3/agent_run.json",
      "cell": "kimi-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.263867,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "hermes",
      "model": "moonshotai/kimi-k3",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 314.765,
      "wave": 3,
      "workload": "eventforge"
    },
    {
      "artifact": "deterministic/eventforge/kimi-openrouter/wave3/openrouter/results/pft/wave3/agent_run.json",
      "cell": "kimi-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.060133,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "pft",
      "model": "moonshotai/kimi-k3",
      "pft_call_diagnostics": {
        "calls": 6,
        "finish_reasons": {
          "ok": 6
        },
        "providers": {
          "BaseTen": 1,
          "Fireworks": 5
        },
        "retry_attempts": 0
      },
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 278.558,
      "wave": 3,
      "workload": "eventforge"
    },
    {
      "artifact": "deterministic/queuecraft/glm-openrouter/wave1/openrouter/results/hermes/wave1/agent_run.json",
      "cell": "glm-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.323866,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "hermes",
      "model": "z-ai/glm-5.2",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 513.471,
      "wave": 1,
      "workload": "queuecraft"
    },
    {
      "artifact": "deterministic/queuecraft/glm-openrouter/wave1/openrouter/results/pft/wave1/agent_run.json",
      "cell": "glm-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.165116,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "pft",
      "model": "z-ai/glm-5.2",
      "pft_call_diagnostics": {
        "calls": 19,
        "finish_reasons": {
          "ok": 19
        },
        "providers": {
          "AtlasCloud": 19
        },
        "retry_attempts": 0
      },
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 114.527,
      "wave": 1,
      "workload": "queuecraft"
    },
    {
      "artifact": "deterministic/queuecraft/glm-openrouter/wave2/openrouter/results/hermes/wave2/agent_run.json",
      "cell": "glm-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.200992,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "hermes",
      "model": "z-ai/glm-5.2",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 288.692,
      "wave": 2,
      "workload": "queuecraft"
    },
    {
      "artifact": "deterministic/queuecraft/glm-openrouter/wave2/openrouter/results/pft/wave2/agent_run.json",
      "cell": "glm-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.527677,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "pft",
      "model": "z-ai/glm-5.2",
      "pft_call_diagnostics": {
        "calls": 24,
        "finish_reasons": {
          "ok": 24
        },
        "providers": {
          "Alibaba": 7,
          "CoreWeave": 16,
          "Ionstream": 1
        },
        "retry_attempts": 0
      },
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 377.436,
      "wave": 2,
      "workload": "queuecraft"
    },
    {
      "artifact": "deterministic/queuecraft/glm-openrouter/wave3/openrouter/results/hermes/wave3/agent_run.json",
      "cell": "glm-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.498547,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "hermes",
      "model": "z-ai/glm-5.2",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 261.392,
      "wave": 3,
      "workload": "queuecraft"
    },
    {
      "artifact": "deterministic/queuecraft/glm-openrouter/wave3/openrouter/results/pft/wave3/agent_run.json",
      "cell": "glm-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.272193,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "pft",
      "model": "z-ai/glm-5.2",
      "pft_call_diagnostics": {
        "calls": 16,
        "finish_reasons": {
          "ok": 16
        },
        "providers": {
          "CoreWeave": 10,
          "SiliconFlow": 6
        },
        "retry_attempts": 0
      },
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 465.68,
      "wave": 3,
      "workload": "queuecraft"
    },
    {
      "artifact": "deterministic/queuecraft/glm-vercel-current-hermes/wave1/vercel/results/hermes/wave1/agent_run.json",
      "cell": "glm-vercel",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.600463,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "hermes",
      "model": "zai/glm-5.2",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 307.413,
      "wave": 1,
      "workload": "queuecraft"
    },
    {
      "artifact": "deterministic/queuecraft/glm-vercel-current-hermes/wave1/vercel/results/pft/wave1/agent_run.json",
      "cell": "glm-vercel",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.359863,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "pft",
      "model": "zai/glm-5.2",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 134.213,
      "wave": 1,
      "workload": "queuecraft"
    },
    {
      "artifact": "deterministic/queuecraft/glm-vercel-current-hermes/wave2/vercel/results/hermes/wave2/agent_run.json",
      "cell": "glm-vercel",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.566592,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "hermes",
      "model": "zai/glm-5.2",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 304.641,
      "wave": 2,
      "workload": "queuecraft"
    },
    {
      "artifact": "deterministic/queuecraft/glm-vercel-current-hermes/wave2/vercel/results/pft/wave2/agent_run.json",
      "cell": "glm-vercel",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.392805,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "pft",
      "model": "zai/glm-5.2",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 183.981,
      "wave": 2,
      "workload": "queuecraft"
    },
    {
      "artifact": "deterministic/queuecraft/glm-vercel-current-hermes/wave3/vercel/results/hermes/wave3/agent_run.json",
      "cell": "glm-vercel",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.572873,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "hermes",
      "model": "zai/glm-5.2",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 178.317,
      "wave": 3,
      "workload": "queuecraft"
    },
    {
      "artifact": "deterministic/queuecraft/glm-vercel-current-hermes/wave3/vercel/results/pft/wave3/agent_run.json",
      "cell": "glm-vercel",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.295302,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "pft",
      "model": "zai/glm-5.2",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 158.72,
      "wave": 3,
      "workload": "queuecraft"
    },
    {
      "artifact": "deterministic/queuecraft/kimi-openrouter/wave1/openrouter/results/hermes/wave1/agent_run.json",
      "cell": "kimi-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 2.062446,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "hermes",
      "model": "moonshotai/kimi-k3",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 705.385,
      "wave": 1,
      "workload": "queuecraft"
    },
    {
      "artifact": "deterministic/queuecraft/kimi-openrouter/wave1/openrouter/results/pft/wave1/agent_run.json",
      "cell": "kimi-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.202949,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "pft",
      "model": "moonshotai/kimi-k3",
      "pft_call_diagnostics": {
        "calls": 9,
        "finish_reasons": {
          "ok": 9
        },
        "providers": {
          "Moonshot AI": 9
        },
        "retry_attempts": 0
      },
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 143.329,
      "wave": 1,
      "workload": "queuecraft"
    },
    {
      "artifact": "deterministic/queuecraft/kimi-openrouter/wave2/openrouter/results/hermes/wave2/agent_run.json",
      "cell": "kimi-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 1.191236,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "hermes",
      "model": "moonshotai/kimi-k3",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 271.082,
      "wave": 2,
      "workload": "queuecraft"
    },
    {
      "artifact": "deterministic/queuecraft/kimi-openrouter/wave2/openrouter/results/pft/wave2/agent_run.json",
      "cell": "kimi-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.34042,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "pft",
      "model": "moonshotai/kimi-k3",
      "pft_call_diagnostics": {
        "calls": 11,
        "finish_reasons": {
          "ok": 11
        },
        "providers": {
          "Fireworks": 5,
          "Moonshot AI": 6
        },
        "retry_attempts": 0
      },
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 119.531,
      "wave": 2,
      "workload": "queuecraft"
    },
    {
      "artifact": "deterministic/queuecraft/kimi-openrouter/wave3/openrouter/results/hermes/wave3/agent_run.json",
      "cell": "kimi-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 1.549447,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "hermes",
      "model": "moonshotai/kimi-k3",
      "pft_call_diagnostics": null,
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 868.005,
      "wave": 3,
      "workload": "queuecraft"
    },
    {
      "artifact": "deterministic/queuecraft/kimi-openrouter/wave3/openrouter/results/pft/wave3/agent_run.json",
      "cell": "kimi-openrouter",
      "conformance_reasons": [],
      "conformance_valid": true,
      "contestant_plus_image_output_estimate_upper_bound_usd": null,
      "contestant_plus_image_output_estimate_usd": null,
      "cost_usd": 0.371941,
      "final_image_count": 0,
      "image_cost_evidence": null,
      "image_count": 0,
      "image_output_estimate_upper_bound_usd": null,
      "image_output_estimate_usd": null,
      "lane": "pft",
      "model": "moonshotai/kimi-k3",
      "pft_call_diagnostics": {
        "calls": 12,
        "finish_reasons": {
          "ok": 12
        },
        "providers": {
          "Fireworks": 6,
          "Moonshot AI": 6
        },
        "retry_attempts": 0
      },
      "returncode": 0,
      "success": true,
      "timed_out": false,
      "uncertain_image_count": 0,
      "wall_seconds": 172.926,
      "wave": 3,
      "workload": "queuecraft"
    }
  ],
  "visual_judgments": [
    {
      "cell": "glm-openrouter",
      "model": "gpt-5.6-sol",
      "passes": [
        {
          "confidence": 0.98,
          "order": {
            "a": "hermes",
            "b": "pft"
          },
          "pass": "normal",
          "response_id": "resp_097873a303841d2f016a68849e692081a1a29e7cf1ec1b069d",
          "site_a_total": 81,
          "site_b_total": 97,
          "underlying_winner": "pft",
          "usage": {
            "input_tokens": 31575,
            "input_tokens_details": {
              "cache_write_tokens": 31572,
              "cached_tokens": 0
            },
            "output_tokens": 4619,
            "output_tokens_details": {
              "reasoning_tokens": 3505
            },
            "total_tokens": 36194
          }
        },
        {
          "confidence": 0.97,
          "order": {
            "a": "pft",
            "b": "hermes"
          },
          "pass": "swapped",
          "response_id": "resp_099bf309fe88488e016a6884f8786c81a3a24b0edcd37c8f3f",
          "site_a_total": 95,
          "site_b_total": 80,
          "underlying_winner": "pft",
          "usage": {
            "input_tokens": 31575,
            "input_tokens_details": {
              "cache_write_tokens": 31572,
              "cached_tokens": 0
            },
            "output_tokens": 5510,
            "output_tokens_details": {
              "reasoning_tokens": 4248
            },
            "total_tokens": 37085
          }
        }
      ],
      "verdict": "pft",
      "verdict_reason": "both balanced A/B orders selected the same underlying site",
      "wave": 1
    },
    {
      "cell": "glm-openrouter",
      "model": "gpt-5.6-sol",
      "passes": [
        {
          "confidence": 0.95,
          "order": {
            "a": "hermes",
            "b": "pft"
          },
          "pass": "normal",
          "response_id": "resp_02f4b76f1900fdae016a68856a5c14819dbc726b9f2afd45a9",
          "site_a_total": 94,
          "site_b_total": 85,
          "underlying_winner": "hermes",
          "usage": {
            "input_tokens": 34437,
            "input_tokens_details": {
              "cache_write_tokens": 34434,
              "cached_tokens": 0
            },
            "output_tokens": 4249,
            "output_tokens_details": {
              "reasoning_tokens": 3106
            },
            "total_tokens": 38686
          }
        },
        {
          "confidence": 0.9,
          "order": {
            "a": "pft",
            "b": "hermes"
          },
          "pass": "swapped",
          "response_id": "resp_04200e66a12eb497016a6885c466b0819e80e7e5c199ce0678",
          "site_a_total": 95,
          "site_b_total": 90,
          "underlying_winner": "pft",
          "usage": {
            "input_tokens": 34437,
            "input_tokens_details": {
              "cache_write_tokens": 34434,
              "cached_tokens": 0
            },
            "output_tokens": 3586,
            "output_tokens_details": {
              "reasoning_tokens": 2378
            },
            "total_tokens": 38023
          }
        }
      ],
      "verdict": "tie_inconclusive_order_sensitive",
      "verdict_reason": "the balanced A/B orders did not select the same underlying site",
      "wave": 2
    },
    {
      "cell": "glm-openrouter",
      "model": "gpt-5.6-sol",
      "passes": [
        {
          "confidence": 0.97,
          "order": {
            "a": "pft",
            "b": "hermes"
          },
          "pass": "normal",
          "response_id": "resp_06e7d8151aff76f5016a688607a2c881a28fa3c54ec89ee566",
          "site_a_total": 93,
          "site_b_total": 84,
          "underlying_winner": "pft",
          "usage": {
            "input_tokens": 33141,
            "input_tokens_details": {
              "cache_write_tokens": 33138,
              "cached_tokens": 0
            },
            "output_tokens": 3572,
            "output_tokens_details": {
              "reasoning_tokens": 2588
            },
            "total_tokens": 36713
          }
        },
        {
          "confidence": 0.93,
          "order": {
            "a": "hermes",
            "b": "pft"
          },
          "pass": "swapped",
          "response_id": "resp_04f91257ce3a8bae016a68864ec60c819f8be4d52fb64e7db9",
          "site_a_total": 87,
          "site_b_total": 94,
          "underlying_winner": "pft",
          "usage": {
            "input_tokens": 33141,
            "input_tokens_details": {
              "cache_write_tokens": 33138,
              "cached_tokens": 0
            },
            "output_tokens": 4768,
            "output_tokens_details": {
              "reasoning_tokens": 3587
            },
            "total_tokens": 37909
          }
        }
      ],
      "verdict": "pft",
      "verdict_reason": "both balanced A/B orders selected the same underlying site",
      "wave": 3
    },
    {
      "cell": "kimi-openrouter",
      "model": "gpt-5.6-sol",
      "passes": [
        {
          "confidence": 0.97,
          "order": {
            "a": "hermes",
            "b": "pft"
          },
          "pass": "normal",
          "response_id": "resp_0c2c99f3ba27bc1c016a6882d48dbc81a288f46c257b110001",
          "site_a_total": 90,
          "site_b_total": 98,
          "underlying_winner": "pft",
          "usage": {
            "input_tokens": 29793,
            "input_tokens_details": {
              "cache_write_tokens": 29790,
              "cached_tokens": 0
            },
            "output_tokens": 3127,
            "output_tokens_details": {
              "reasoning_tokens": 2070
            },
            "total_tokens": 32920
          }
        },
        {
          "confidence": 0.95,
          "order": {
            "a": "pft",
            "b": "hermes"
          },
          "pass": "swapped",
          "response_id": "resp_058c66903f74b8f4016a688315b67c8191b610073acbeca3b8",
          "site_a_total": 97,
          "site_b_total": 90,
          "underlying_winner": "pft",
          "usage": {
            "input_tokens": 29793,
            "input_tokens_details": {
              "cache_write_tokens": 29790,
              "cached_tokens": 0
            },
            "output_tokens": 3500,
            "output_tokens_details": {
              "reasoning_tokens": 2554
            },
            "total_tokens": 33293
          }
        }
      ],
      "verdict": "pft",
      "verdict_reason": "both balanced A/B orders selected the same underlying site",
      "wave": 1
    },
    {
      "cell": "kimi-openrouter",
      "model": "gpt-5.6-sol",
      "passes": [
        {
          "confidence": 0.88,
          "order": {
            "a": "hermes",
            "b": "pft"
          },
          "pass": "normal",
          "response_id": "resp_09bee529677f6e61016a68835a81a48191aa8a4ce383ca67cf",
          "site_a_total": 92,
          "site_b_total": 96,
          "underlying_winner": "pft",
          "usage": {
            "input_tokens": 31467,
            "input_tokens_details": {
              "cache_write_tokens": 31464,
              "cached_tokens": 0
            },
            "output_tokens": 3654,
            "output_tokens_details": {
              "reasoning_tokens": 2342
            },
            "total_tokens": 35121
          }
        },
        {
          "confidence": 0.84,
          "order": {
            "a": "pft",
            "b": "hermes"
          },
          "pass": "swapped",
          "response_id": "resp_00e7c2a5b07dcf15016a6883a03ce881a2a41ce13291f428bb",
          "site_a_total": 96,
          "site_b_total": 92,
          "underlying_winner": "pft",
          "usage": {
            "input_tokens": 31467,
            "input_tokens_details": {
              "cache_write_tokens": 31464,
              "cached_tokens": 0
            },
            "output_tokens": 4139,
            "output_tokens_details": {
              "reasoning_tokens": 2978
            },
            "total_tokens": 35606
          }
        }
      ],
      "verdict": "pft",
      "verdict_reason": "both balanced A/B orders selected the same underlying site",
      "wave": 2
    },
    {
      "cell": "kimi-openrouter",
      "model": "gpt-5.6-sol",
      "passes": [
        {
          "confidence": 0.92,
          "order": {
            "a": "hermes",
            "b": "pft"
          },
          "pass": "normal",
          "response_id": "resp_044eaa94865f9638016a6883efd628819c858f0184f826199c",
          "site_a_total": 90,
          "site_b_total": 94,
          "underlying_winner": "pft",
          "usage": {
            "input_tokens": 31251,
            "input_tokens_details": {
              "cache_write_tokens": 31248,
              "cached_tokens": 0
            },
            "output_tokens": 4810,
            "output_tokens_details": {
              "reasoning_tokens": 3506
            },
            "total_tokens": 36061
          }
        },
        {
          "confidence": 0.9,
          "order": {
            "a": "pft",
            "b": "hermes"
          },
          "pass": "swapped",
          "response_id": "resp_09b50af7a87e7226016a688453036c81a09e7a41f9161c9aae",
          "site_a_total": 94,
          "site_b_total": 88,
          "underlying_winner": "pft",
          "usage": {
            "input_tokens": 31251,
            "input_tokens_details": {
              "cache_write_tokens": 31248,
              "cached_tokens": 0
            },
            "output_tokens": 3739,
            "output_tokens_details": {
              "reasoning_tokens": 2433
            },
            "total_tokens": 34990
          }
        }
      ],
      "verdict": "pft",
      "verdict_reason": "both balanced A/B orders selected the same underlying site",
      "wave": 3
    },
    {
      "cell": "opus",
      "model": "gpt-5.6-sol",
      "passes": [
        {
          "confidence": 0.95,
          "order": {
            "a": "cc",
            "b": "pft"
          },
          "pass": "normal",
          "response_id": "resp_04a2bf03abdcc12c016a681d41e7ac819d93e2fadebc253660",
          "site_a_total": 94,
          "site_b_total": 90,
          "underlying_winner": "cc",
          "usage": {
            "input_tokens": 38703,
            "input_tokens_details": {
              "cache_write_tokens": 38700,
              "cached_tokens": 0
            },
            "output_tokens": 4085,
            "output_tokens_details": {
              "reasoning_tokens": 2921
            },
            "total_tokens": 42788
          }
        },
        {
          "confidence": 0.94,
          "order": {
            "a": "pft",
            "b": "cc"
          },
          "pass": "swapped",
          "response_id": "resp_0e996b0759b36842016a681d86fb60819f9fd673a6adb6cecb",
          "site_a_total": 91,
          "site_b_total": 97,
          "underlying_winner": "cc",
          "usage": {
            "input_tokens": 38703,
            "input_tokens_details": {
              "cache_write_tokens": 38700,
              "cached_tokens": 0
            },
            "output_tokens": 4167,
            "output_tokens_details": {
              "reasoning_tokens": 2995
            },
            "total_tokens": 42870
          }
        }
      ],
      "verdict": "cc",
      "verdict_reason": "both balanced A/B orders selected the same underlying site",
      "wave": 1
    },
    {
      "cell": "opus",
      "model": "gpt-5.6-sol",
      "passes": [
        {
          "confidence": 0.94,
          "order": {
            "a": "cc",
            "b": "pft"
          },
          "pass": "normal",
          "response_id": "resp_04949cbcc4b7c72b016a681dd1b4bc81a3957c183d81374832",
          "site_a_total": 89,
          "site_b_total": 95,
          "underlying_winner": "pft",
          "usage": {
            "input_tokens": 38001,
            "input_tokens_details": {
              "cache_write_tokens": 37998,
              "cached_tokens": 0
            },
            "output_tokens": 4050,
            "output_tokens_details": {
              "reasoning_tokens": 3106
            },
            "total_tokens": 42051
          }
        },
        {
          "confidence": 0.96,
          "order": {
            "a": "pft",
            "b": "cc"
          },
          "pass": "swapped",
          "response_id": "resp_09ead2c455dafb03016a681e1d7dd0819d8e4316196190292d",
          "site_a_total": 95,
          "site_b_total": 88,
          "underlying_winner": "pft",
          "usage": {
            "input_tokens": 38001,
            "input_tokens_details": {
              "cache_write_tokens": 37998,
              "cached_tokens": 0
            },
            "output_tokens": 3225,
            "output_tokens_details": {
              "reasoning_tokens": 1999
            },
            "total_tokens": 41226
          }
        }
      ],
      "verdict": "pft",
      "verdict_reason": "both balanced A/B orders selected the same underlying site",
      "wave": 2
    },
    {
      "cell": "opus",
      "model": "gpt-5.6-sol",
      "passes": [
        {
          "confidence": 0.96,
          "order": {
            "a": "cc",
            "b": "pft"
          },
          "pass": "normal",
          "response_id": "resp_08d806ead87241bc016a681e5c0cfc81a285668614d0e28ad3",
          "site_a_total": 97,
          "site_b_total": 97,
          "underlying_winner": "tie",
          "usage": {
            "input_tokens": 40971,
            "input_tokens_details": {
              "cache_write_tokens": 40968,
              "cached_tokens": 0
            },
            "output_tokens": 5534,
            "output_tokens_details": {
              "reasoning_tokens": 4469
            },
            "total_tokens": 46505
          }
        },
        {
          "confidence": 0.91,
          "order": {
            "a": "pft",
            "b": "cc"
          },
          "pass": "swapped",
          "response_id": "resp_0c64dac8f47e4d07016a681ebed364819296de2b4cbe22b939",
          "site_a_total": 96,
          "site_b_total": 93,
          "underlying_winner": "pft",
          "usage": {
            "input_tokens": 40971,
            "input_tokens_details": {
              "cache_write_tokens": 40968,
              "cached_tokens": 0
            },
            "output_tokens": 4691,
            "output_tokens_details": {
              "reasoning_tokens": 3403
            },
            "total_tokens": 45662
          }
        }
      ],
      "verdict": "tie_inconclusive_order_sensitive",
      "verdict_reason": "the balanced A/B orders did not select the same underlying site",
      "wave": 3
    }
  ],
  "visual_verdict_summary": [
    {
      "cell": "glm-openrouter",
      "inconclusive": 1,
      "opponent": "hermes",
      "opponent_wins": 0,
      "pft_wins": 2,
      "waves": 3
    },
    {
      "cell": "kimi-openrouter",
      "inconclusive": 0,
      "opponent": "hermes",
      "opponent_wins": 0,
      "pft_wins": 3,
      "waves": 3
    },
    {
      "cell": "opus",
      "inconclusive": 1,
      "opponent": "cc",
      "opponent_wins": 1,
      "pft_wins": 1,
      "waves": 3
    }
  ]
}
